xturing: A Python repository from psiberx

Build and control your own LLMs

xturing provides fast, efficient and simple fine-tuning of LLMs, such as LLaMA, GPT-J, GPT-2, OPT, Cerebras-GPT, Galactica, and more. By providing an easy-to-use interface for personalizing LLMs to your own data and application, xTuring makes it simple to build and control LLMs. The entire process can be done inside your computer or in your private cloud, ensuring data privacy and security.

With xturing you can,

Ingest data from different sources and preprocess them to a format LLMs can understand
Scale from single to multiple GPUs for faster fine-tuning
Leverage memory-efficient techniques (i.e. INT4, LoRA fine-tuning) to reduce your hardware costs by up to 90% of the time
Explore different fine-tuning methods and benchmark them to find the best performing model
Evaluate fine-tuned models on well-defined metrics for in-depth analysis

🌟 New feature - INT4 fine-tuning with LLaMA LoRA

We are excited to announce the latest enhancement to our xTuring library: INT4 fine-tuning demo. With this update, you can fine-tune LLMs like LLaMA with LoRA architecture in INT4 precision with less than 6GB of VRAM. This breakthrough significantly reduces memory requirements and accelerates the fine-tuning process, allowing you to achieve state-of-the-art performance with less computational resources.

More information about INT4 fine-tuning and benchmarks can be found in the INT4 README.

You can check out the LLaMA INT4 fine-tuning example to see how it works.

CLI playground

UI playground

⚙️ Installation

pip install xturing

🚀 Quickstart

from xturing.datasets import InstructionDataset
from xturing.models import BaseModel

# Load the dataset
instruction_dataset = InstructionDataset("./alpaca_data")

# Initialize the model
model = BaseModel.create("llama_lora")

# Finetune the model
model.finetune(dataset=instruction_dataset)

# Perform inference
output = model.generate(texts=["Why LLM models are becoming so important?"])

print("Generated output by the model: {}".format(output))

You can find the data folder here.

📚 Tutorials

📊 Performance

Here is a comparison for the performance of different fine-tuning techniques on the LLaMA 7B model. We use the Alpaca dataset for fine-tuning. The dataset contains 52K instructions.

Hardware:

4xA100 40GB GPU, 335GB CPU RAM

Fine-tuning parameters:

{
  'maximum sequence length': 512,
  'batch size': 1,
}

LLaMA 7B	DeepSpeed + CPU Offloading	LoRA + DeepSpeed	LoRA + DeepSpeed + CPU Offloading
GPU	33.5 GB	23.7 GB	21.9 GB
CPU	190 GB	10.2 GB	14.9 GB
Time per epoch	21 hours	20 mins	20 mins

Please submit your performance results on other GPUs.

📎 Fine-tuned model checkpoints

We have already fine-tuned some models that you can use as your base or start playing with. Here is how you would load them:

from xturing.models import BaseModel
model = BaseModel.load("x/distilgpt2_lora_finetuned_alpaca")

model	dataset	Path
DistilGPT-2 LoRA	alpaca	`x/distilgpt2_lora_finetuned_alpaca`
LLaMA LoRA	alpaca	`x/llama_lora_finetuned_alpaca`

📈 Roadmap

Support for LLaMA, GPT-J, GPT-2, OPT, Cerebras-GPT, Galactica and Bloom models
Dataset generation using self-instruction
2x more memory-efficient fine-tuning vs LoRA and unsupervised fine-tuning
INT8 low-precision fine-tuning support
Supports OpenAI, Cohere and AI21 Studio model APIs for dataset generation
Added fine-tuned checkpoints for some models to the hub
INT4 LLaMA LoRA fine-tuning demo
Evaluation of LLM models
Support for Stable Diffusion

🤝 Help and Support

If you have any questions, you can create an issue on this repository.

You can also join our Discord server and start a discussion in the #xturing channel.

📝 License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

🌎 Contributing

As an open source project in a rapidly evolving field, we welcome contributions of all kinds, including new features and better documentation. Please read our contributing guide to learn how you can get involved.

psiberx/xturing